MoDern: A Sparse Model-inspired Deep Thresholding Network for Exponential Signal Reconstruction—Application in Fast Biological Spectroscopy
MoDern is a reliable, widely-available, ultra-fast, and easy-to-use technique for highly accelerated NMR. This work develops the first cloud-based artificial intelligence platform (XCloud-MoDern) for multi-dimensional NMR data processing, and is the proof-of-concept demonstration of the effectiveness of merging optimization, deep learning, and cloud computing.
The preprint paper can be seen in https://arxiv.org/abs/2012.14830.
This paper has been accepted by IEEE Transactions on Neural Networks and Learning Systems (2022) in http://dx.doi.org/10.1109/TNNLS.2022.3144580.
Email: Xiaobo Qu (quxiaobo@xmu.edu.cn) CC: Zi Wang (wangzi1023@stu.xmu.edu.cn)
Homepage: http://csrc.xmu.edu.cn
XCloud-MoDern is an easy-to-use cloud computing platform for processing of non-uniformly sampled (NUS) multi-dimensional NMR spectra online. Up to now, XCloud-MoDern uses model-inspired deep learning (MoDern) to fast recover high-quality multi-dimensional spectra from NUS data. The platform also provides a customized retrospectively undersampling technique (NUS simulator) to produce NUS data and the corresponding NUS mask from the fully sampled NMR data.
Now, XCloud-MoDern is available at: http://36.138.17.102:8989/, and we will continue to improve the using feeling.
Details on the instructions of XCloud-MoDern are described in its Manual, you can download "Manual_XCloud-MoDern.pdf" here (old version) or on our cloud platform (latest version).
We also have provided some demo data and scripts on the cloud for the quick try, you can download "Demo_data_scripts_XCloud-MoDern.zip" here (old version) or on our cloud platform (latest version).
Hope you enjoy the reliable, efficient, and high-performance experience. If you find any questions in using XCloud-MoDern, please email me at wangzi1023@stu.xmu.edu.cn.
The synthetic training datasets used in MoDern are shared at: https://drive.google.com/file/d/1bZAP-ittB94wm0hB3SfVsvqXGwk05tkg/view?usp=sharing.
After requesting the access, please email me at wangzi1023@stu.xmu.edu.cn.
If you want to use the platform and training datasets, please cite the following paper:
Zi Wang et al., A Sparse Model-inspired Deep Thresholding Network for Exponential Signal Reconstruction—Application in Fast Biological Spectroscopy, IEEE Transactions on Neural Networks and Learning Systems, 34(10): 7578-7592, 2023.